Good work! Glad you got what you needed Ben.
My take on MVVM really comes down to responsibility and separation of concerns. The view is about receiving input and setting how to display information (map click, formatting and displaying the coordinates, etc...). The view model is about retrieving data from the model or view, performing some presentation logic (publishing the coordinates via Messenger or whatever else you would like to do with them), and being able to test that logic. The Action in this case is just a tool to better wire those two together. Because its an Action, there's also the added benefit of being reusable and blendable.
In my opinion, lots of MVVM newbies get caught up with the no code behind nonsense. In this case, you could have just as easily added a handler for the Map control's MouseMove event and set the property on the View Model in the code behind directly. You would still have the same separation of concerns and still just as testable. Either way, the coordinate transformation logic and wiring to the view model is handled by the view. Without the use of an Action, you would have sacrificed the reuse and blendability, but maybe that's not a big deal for your situation.
... just my .02